Deletion of OTX2 in neural ectoderm delays anterior pituitary development

Abstract

OTX2 is a homeodomain transcription factor that is necessary for normal head development in mouse and man. Heterozygosity for loss-of-function alleles causes an incompletely penetrant, haploinsufficiency disorder. Affected individuals exhibit a spectrum of features that range from developmental defects in eye and/or pituitary development to acephaly. To investigate the mechanism underlying the pituitary defects, we used different cre lines to inactivate Otx2 in early head development and in the prospective anterior and posterior lobes. Mice homozygous for Otx2 deficiency in early head development and pituitary oral ectoderm exhibit craniofacial defects and pituitary gland dysmorphology, but normal pituitary cell specification. The morphological defects mimic those observed in humans and mice with OTX2 heterozygous mutations. Mice homozygous for Otx2 deficiency in the pituitary neural ectoderm exhibited altered patterning of gene expression and ablation of FGF signaling. The posterior pituitary lobe and stalk, which normally arise from neural ectoderm, were extremely hypoplastic. Otx2 expression was intact in Rathke's pouch, the precursor to the anterior lobe, but the anterior lobe was hypoplastic. The lack of FGF signaling from the neural ectoderm was sufficient to impair anterior lobe growth, but not the differentiation of hormone-producing cells. This study demonstrates that Otx2 expression in the neural ectoderm is important intrinsically for the development of the posterior lobe and pituitary stalk, and it has significant extrinsic effects on anterior pituitary growth. Otx2 expression early in head development is important for establishing normal craniofacial features including development of the brain, eyes and pituitary gland.

INTRODUCTION

Patients heterozygous for mutations in the transcription factor OTX2 exhibit a spectrum of phenotypes that can include severe ocular defects, central nervous system abnormalities, developmental delay, endocrine deficiencies, and/or structural and functional abnormalities of the pituitary gland (1–6). The pituitary gland defects, if present, are also variable and can include isolated growth hormone deficiency (IGHD), combined pituitary hormone deficiency (CPHD), hypogonadotropic hypogonadism, anterior pituitary hypoplasia, ectopic neurohypophysis (posterior pituitary lobe) and disrupted pituitary stalk [reviewed (7)]. The majority of the reported mutations are nonsense mutations or frameshifts that result in truncated proteins (1,3–6,8). Functional studies demonstrate that truncation of the OTX2 protein leads to reduced transactivation activity, although some mutations may act in a dominant-negative manner (2,3,5). No clear genotype–phenotype correlation is associated with these mutations, and phenotypic variability occurs even among individuals with the same mutation (2,5,8). Furthermore, OTX2 mutations that have been examined in families all exhibit incomplete penetrance, which may be due to the influence of modifying effects of other genes or epigenetic events (9). In one typical example, a patient heterozygous for an OTX2 mutation presented with short stature, IGHD, pituitary hypoplasia, ectopic posterior pituitary gland and anophthalmia. The father carried the same mutation, and although he had short stature (<5th percentile), he did not have any ocular or pituitary defects (10).

Studies in mice support the idea that the incomplete penetrance of OTX2 loss-of-function mutations is caused by modifier genes that enhance or suppress the phenotype. Homozygous deletion of Otx2 consistently causes embryonic lethality in mice because they lack the rostral neural ectoderm, from which the forebrain, midbrain and rostral hindbrain arise (11). Heterozygous mutant mice have a variable phenotype that resembles aspects of the OTX2 features in human patients. The phenotypic range spans from unaffected, to craniofacial malformations affecting the eyes (anophthalmia and microphthalmia), jaw (agnathia and micrognathia) and nose, and, finally, to severe developmental defects of the head (acephaly and holoprosencephaly) (12). The pituitary phenotypes extend from deformities of the posterior pituitary lobe and dysmorphology of the anterior lobe (adenohypophysis) to pituitary aplasia. The genetic background of the mice affects the severity and frequency of all of the phenotypes, C57BL/6 enhances the phenotype of heterozygous mutants and CBA suppresses it. Genetic mapping revealed the presence of two interacting loci that underlie this variation, but the genes are unknown (13).

OTX2 has been proposed to influence pituitary development by regulating expression of two transcription factors that are important for anterior pituitary development: HESX1 and POU1F1 (4,5,10). For a review of pituitary organogenesis, see Davis et al. (2009) (14). If OTX2 was a direct, master regulator of either of these genes, there would be overlap in the expression patterns. The temporal and spatial expression of these three transcription factors and their loss-of-function phenotypes suggest a different and complex underlying mechanism, however.

The correspondence between Otx2 and Hesx1 expression in early embryogenesis supports the idea that OTX2 regulates Hesx1 in developing anterior structures (15,16). Both genes are expressed prior to implantation, and at gastrulation, their expression becomes localized in the anterior visceral endoderm (AVE) and later in the neural ectoderm, respectively (17–22). In addition, both genes are expressed in the forebrain. There are many similarities in loss-of-function phenotypes of Hesx1 and Otx2 mutants, including development of the head, eyes, brain and pituitary gland (23,24). This is consistent with the idea that Otx2 regulates Hesx1 in early head development.

During pituitary development, the expression of Otx2 is broader than Hesx1. Otx2 is expressed at e10.5 in the ventral diencephalon, which will become the pituitary stalk and posterior lobe, and it persists there through e16.5. Hesx1 is not expressed in the ventral diencephalon or posterior lobe. Both Otx2 and Hesx1 are expressed modestly in Rathke's pouch, which eventually develops into the intermediate and anterior pituitary lobes. Otx2 expression is silenced by e12.5, and Hesx1 is by e14.5 (19,22,25). Thus, it is possible that Otx2 regulates Hesx1 in the oral ectoderm that produces the pouch, and in early head development, but not in the neural ectoderm that becomes the posterior lobe.

There is no overlap between Otx2 and Pou1f1 expression in the pituitary primordium or anterior lobe. Pou1f1 is expressed almost exclusively in the developing anterior lobe, and transcripts are not detectable until e14.5, well after Otx2 and Hesx1 expression is silenced (26). In addition, Pou1f1 mutations only affect the differentiation of three pituitary hormone-producing cell types leading to a triad of pituitary hormone deficiencies: thyrotropin (TSH), growth hormone (GH) and prolactin (27–31). If there is a role for Otx2 in Rathke's pouch, it does not involve direct regulation of Pou1f1.

We hypothesize that OTX2 promotes anterior pituitary development by activating target genes in the neural ectoderm. To test this idea, we generated conditional deletions of Otx2 in the neural ectoderm and in the oral ectoderm. Our data with the conditional oral ectoderm knockouts suggest that Otx2 plays a minor role, if any, in the organogenesis of Rathke's pouch. We discovered that Otx2 expression in the neural ectoderm, however, is necessary for normal development of the infundibulum or pituitary stalk, and for initial induction of FGF signaling in the ventral diencephalon. FGF signaling is known to have a crucial role in anterior pituitary gland growth and development (32). Thus, Otx2 deficiency in the neural ectoderm is sufficient to delay normal growth and development of the anterior pituitary gland.

RESULTS

Conditional Otx2 oral ectoderm knock out

We sought to remove Otx2 in the prospective anterior lobe of the pituitary gland using the Foxg1-cre mouse line (33). Foxg1-cre expression is reported to be detectable in every cell in the oral ectoderm that gives rise to the anterior and intermediate lobes of the pituitary, before the closure of Rahtke's pouch at e9.5. There is no expression of Foxg1-cre in the ventral diencephalon (34). We mated Otx2^FX/+; Foxg1-cre mice to Otx2^FX/FX mice to generate Otx2^FX/FX; Foxg1-cre mice which were designated as mutants (Otx2^FX;Foxg1-cre) and Otx2^FX/FX and Otx2^FX/+ mice were designated as controls.

Otx2^FX;Foxg1-cre embryos collected at e14.5 had obvious craniofacial deformities that mimic those found in OTX2 heterozygote humans and mice (12,35,36). At e12.5 and e14.5, the defects in Otx2^FX;Foxg1-cre mutants include microphthalmia or anophthalmia, facial defects and/or excencephaly (n = 7) (Fig. 1A and C). This indicates inactivation of Otx2 early during head development, and a lack of anterior pituitary specificity. The pituitary glands of the mutants exhibited a variety of phenotypes that include dysmorphology, protrusion through the palate, misplacement of a hypoplastic pituitary in the head or a relatively normal looking pituitary (n = 6). All the Otx2^FX;Foxg1-cre pituitaries express the pituitary transcription factors PITX1 at e14.5 and Hesx1 at e11.5. Immunostaining of the mutant anterior pituitaries revealed normal expression of the signature transcription factors POU1F1 (somatotropes, lactotropes and thyrotropes), TPIT (corticotropes) and the hormone subunit αGSU, which marks gonadotropes and thyrotropes, at e14.5 (Fig. 1B). These data suggest that the morphological defects characteristic of the Otx2^FX;Foxg1-cre pituitaries are due to loss of OTX2 during early head development, whereas loss of OTX2 expression in Rathke's pouch does not affect cell specification.

Figure 1.

(A–C) Otx2^FX;Foxg1-cre-mutant embryos have craniofacial defects and dysmorphic pituitary glands. (A) At e14.5, Otx2^FX;Foxg1-cre-mutant embryos exhibit craniofacial (top middle panel) and ocular defects (top right panel). Hematoxylin and eosin staining of e14.5 sagittal sections of Otx2^FX;Foxg1-cre mutants reveal dysmorphic pituitary glands (arrow in middle panel and right panels). A—anterior pituitary lobe, I—intermediate pituitary lobe, P—posterior pituitary lobe. Photos were taken at. Scale bar: 100 μm. Immunostaining for PITX1 reveals which tissues are pituitary specific in the OTX2^FX;Foxg1-cre mutants at e14.5. The photos in the third panel were taken at ×200. Scale bar: 50 μm. In situ hybridization for Hesx1 shows normal expression in smaller Otx2^FX;Foxg1-cre mutant pituitaries at e11.5. RP—Rathke's Pouch, INF—infundibulum. The photos in the bottom panel were taken at. Scale bar: 50 μm. (B). Otx2^FX;Foxg1-cre mutants express POU1F1, TPIT and αGSU at e14.5 when compared with control pituitaries. Expression was variable and dependent on the growth of the mutant anterior lobe. Photos were taken at. Scale bar 50 μm. (C) Chart quantifying external defects observed in Otx2^FX;Foxg1-cre-mutant embryos at e12.5 and e14.5 (N = 7).

We used a different cre line, Pitx2-cre, to conditionally delete Otx2 in the oral ectoderm that gives rise to the pituitary gland. The cre cassette is inserted into exon 5 of the Pitx2 gene (37). The Otx2^FX/FX (38) mouse was crossed with a cre reporter strain consisting of a floxed stop sequence, which upon cre-mediated excision permits lacZ expression (R26R^FX/FX) (39). The Otx2^FX/FX; R26R^FX/FX offspring were mated to Otx2^FX/+; Pitx2-cre^+/− mice. Mice with genotype Otx2^FX/FX; R26R^FX/+; Pitx2-cre^+/− are designated as mutants (Otx2^FX; Pitx2-cre) and Otx2^FX/+; R26R^FX/+ mice serve as normal controls. We assessed the efficiency and penetrance of cre-mediated recombination using X-gal staining of frozen sections from e11.5 embryos (Supplementary Material, Fig. S1A). The mutant embryos consistently had blue X-gal staining in approximately two-thirds of the cells in Rathke's pouch, and no cells were stained in the ventral diencephalon (n = 3). There is no evidence of nonspecific X-gal staining in normal embryos (n = 3). Hematoxylin and eosin staining of sections from e14.5 mutant embryos revealed no morphological difference between the Otx2^FX;Pitx2-cre mutants and controls (Supplementary Material, Fig. S1B). Furthermore, the mutant pituitaries exhibited normal cell specification because e14.5 mutant embryos had normal immunostaining for αGSU and POU1F1 (Supplementary Material, Fig. S1B), and the P10 neonates had normal immunostaining for GH, TSHβ, ACTH and LHβ (data not shown). This is consistent with the idea that the pituitary defects in the Otx2^FX;Foxg1-cre mice result from deletion of Otx2 in early head development, not Rathke's pouch.

Conditional Otx2 neural ectoderm knock out

To generate conditional Otx2 neural ectoderm knockout mice, we mated the Otx2^FX/FX (38) mouse to the Nkx2.1-cre driver strain in which the cre cassette is inserted at the Nkx2.1 locus (40). The Otx2^FX/FX; R26R^FX/FX offspring were then mated to Otx2^FX/+; Nkx2.1-cre^+/− mice. Mice with the genotype Otx2^FX/FX; R26R^FX/+; Nkx2.1-cre^+/− are designated as mutants (Otx2^FX;Nkx2.1-cre), and Otx2^FX/+; R26R^FX/FXmice serve as normal controls. We assessed the efficiency and penetrance of cre-mediated recombination using X-gal staining of frozen sections from e11.5 Otx2^FX;Nkx2.1-cre mutant and control embyros (Fig. 2A). The mutant embryos consistently had blue X-gal staining throughout the ventral diencephalon, but staining in Rathke's pouch was minimal: only 1/7 embryos had X-gal stained cells in Rathke's pouch. There is no evidence of other nonspecific X-gal staining in Otx2^FX;Nkx2.1-cre embryos (n = 7). To verify that Otx2 was effectively knocked out in the ventral diencephalon, we used immunohistochemistry to detect the OTX2 protein at e11.5, e12.5 and e14.5. Almost no OTX2 protein was detected in the ventral diencephalon in the Otx2^FX;Nkx2.1-cre mutants at either time (Fig. 2B) (N = 3 for e11.5, N = 3 for e12.5 and N = 5 for e14.5). OTX2 immunostaining in Rathke's pouch of the mutant is indistinguishable from the wild type, affirming that this transgenic knockout strategy effectively and specifically removed OTX2 expression from the ventral diencephalon.

Figure 2.

(A and B) Nkx2.1-cre activity is specific to the ventral diencephalon and is effective at removing OTX2 expression. (A) Sagittal frozen sections of e11.5 control and Otx2^FX;Nkx2.1cre;RosalacZ^FX/+ mutant embryos were examined with Xgal staining to assess cre activity and effective recombination. NE—neural ectoderm, RP—Rathke's Pouch. Photos were taken at. Scale bar: 100 μm. (B) Immunohistochemistry of OTX2 protein on sagittal sections at e11.5, e12.5 and e14.5 reveal that OTX2 expression is absent from the ventral diencephalon in the Otx2^FX;Nkx2.1-cremutant pituitaries. INF—infundibulum, AP—anterior pituitary. The photos in the top and bottom panels were taken at. Scale bar: 100 μm. The photos in the middle panel were taken at. Scale bar 50 μm.

Neural ectoderm knockout of Otx2 results in poor posterior lobe formation and a smaller anterior pituitary gland

To assess the effects of neural-ectoderm-specific deletion of Otx2, we examined the morphology of Otx2^FX;Nkx2.1-cremutant pituitaries using hematoxylin and eosin staining of sections from embryos collected at e10.5, e12.5, e14.5 and e16.5 (n = 3 or greater for each time point). There was no evidence of invagination of the ventral diencephalon, which is necessary for forming the infundibulum and posterior lobe of the pituitary gland. Indeed, at e14.5 and e16.5, there appears to be no posterior lobe present in sagittal sections (Fig. 3A). There were no distinguishable abnormalities in Rathke's pouch development in Otx2^FX;Nkx2.1-cre mutants at e10.5, when the ventral diencephalon begins to invaginate and the pouch is induced. While Otx2 knockout in the neural ectoderm did not affect the separation of Rathke's pouch from the oral ectoderm, the mutant pouch is obviously hypoplastic at e14.5. The genetic background of the Otx2^FX; Nkx2.1cre is mixed, consisting of C57BL/6J and 129/SV, but there is no obvious phenotypic variability in the pituitaries of these animals.

Figure 3.

(A and B) Loss of OTX2 in the ventral diencephalon results in a smaller, dysmorphic pituitary gland. (A) e10.5, e12.5, e14.5 and e16.5 sagittal pituitary sections were stained using hematoxylin and eosin. At e10.5, there is no difference between Otx2^FX/+wild-type (top panel) and Otx2^FX;Nkx2.1-cre mutants (bottom panel). At e12.5 through e16.5, there is a lack of invagination of the infundibulum and a smaller anterior lobe in the mutants compared with the wild types. The e10.5 and e14.5 photos were taken at ×200. Scale bar: 50 μm. The e12.5 and e16.5 photos were taken at ×100. Scale bar: 100 μm. (B) e14.5 coronal pituitary sections were stained using hematoxylin and eosin, revealing a slight invagination of ventral diencephalon in the mutants (black arrow). TLE4 immunostaining at e14.5 identifies the reduced evagination of the posterior lobe in the Otx2^FX;Nkx2.1-cre mutants (white arrow). OTX2 immunostaining in control and Otx2^FX;Nkx2.1-cre-mutant animals at e14.5 reveals some OTX2 protein in the small mutant posterior lobe (white arrowhead.) In situ hybridization for Otx2 in the control and Otx2^FX;Nkx2.1-cre mutants reveals some transcripts in the small, mutant posterior lobe (black arrow head). III—third ventricle, A—anterior lobe, P—posterior lobe. Photos were taken at. Scale bar: 50 μm.

We investigated further by staining serial coronal pituitary sections of e14.5 Otx2^FX;Nkx2.1-cre mutants and normal littermates with hematoxylin and eosin. A nubbin of tissue resembling the posterior lobe was noted at 100% penetrance and in at least three sections per animal. (Fig. 3B). To determine whether this tissue was specified as posterior lobe, we carried out immunostaining for the co-repressor TLE4 (transducin-like enhancer of split), which is normally concentrated in the developing posterior lobe at e14.5 (41). The nubbin of tissue in the Otx2^FX;Nkx2.1-cremutant animals expressed TLE4, indicating commitment to the posterior lobe fate (Fig. 3, n = 3 per genotype). We also examined OTX2 immunostaining in Otx2^FX;Nkx2.1-cre mutants at e14.5 and found minimal OTX2 staining in the posterior lobe tissue, indicating efficient, specific elimination of OTX2 (Fig. 3B, n = 4). Occasionally, a few cells with OTX2 expression were detected in posterior pituitary tissue. In situ hybridization with the Otx2 in situ probe mimicked the results seen with the OTX2 antibody (Fig. 3B, n = 3). These data indicate that the nubbin of posterior pituitary tissue present in the mutants is unlikely to have arisen from cells in the ventral diencephalon that escaped Otx2 deletion.

At postnatal day 2 (P2), both the posterior and anterior lobes of Otx2^FX;Nkx2.1-cre mutants are significantly smaller than normal (Fig. 4A). There were significantly fewer sections with pituitary tissue in embryos collected at either e14.5 or P2 (Fig. 4B). The overall reduction in pituitary tissue was comparable at e14.5 and P2, 41 and 40, respectively. The size reduction was similar in the P2 anterior and posterior lobes: 52 and 38%, respectively (Fig. 4C). Surprisingly, improved mutant pituitary growth is evident at P10 (Fig. 4A). There were significantly fewer sections of Otx2^FX;Nkx2.1 cre mutant pituitary tissue at P10, but the size was only reduced by 22% (Fig. 4B). These results suggest that the primary effect of knocking OTX2 out of the ventral diencephalon is a delay in posterior lobe formation, and secondary to that, the size of Rathke's pouch is decreased.

Figure 4.

(A and B) Delayed invagination of the posterior lobe in Otx2^FX;Nkx2.1-cre mutants. (A) Coronal sections at P2 and P10 were hematoxylin and eosin stained. Photos were taken at. Scale bar: 100 μm. (B) Relative pituitary size was measured at e14.5, P2 and P10. Three control and three mutant animals were sectioned and mounted two sections per slide. Sections containing pituitary gland tissue were counted and averaged, and percentages were calculated. P-values for control (gray bars) compared with mutants (black bars) were determined using the average number of sections and the Student's T-test for each age group: P-value < 0.05 (*). (C) ImageJ IJ 1.46r software from the NIH was used to measure the posterior and anterior lobes of three control and three mutants with four sections per sample. The Student's T-test was used to calculate P-values for controls compared with Nkx2.1cre^+/− mutants for each age group: P-value < 0.05 (*) for posterior lobe; P-value < 0.01 (**) for anterior lobe.

FGF signaling is disrupted in the Otx2^FX; Nkx2.1-cre knockout pituitary

We tested the hypothesis that Otx2 expression in the neural ectoderm is necessary for FGF signaling, which in turn is required for stimulating anterior pituitary development. FGF signaling normally emanates from the ventral diencephalon from e9.0 to e14.5 (16), and deficiency of FGF10 (32) or FGFr2b (42) is sufficient to cause failed anterior pituitary development. We examined FGF signaling in Otx2^FX; Nkx2.1-cre mutants by performing in situ hybridization for Fgf10 at e12.5. We found that Fgf10 expression is greatly decreased in the Otx2^FX; Nkx2.1-cre mutant ventral diencephalon compared with normal controls (Fig. 5A). Multiple FGFs are expressed in the ventral diencephalon, and they all should increase the phosphorylation of ERK (pERK). We carried out immunostaining for pERK in sections from e11.5 embryos. There is an obvious decrease in pERK immunostaining in the ventral diencephalon of the Otx2^FX;Nkx2.1-cre mutant relative to normal controls (Fig. 5A), which is consistent with an overall deficiency in FGF signaling. These results suggest that the delay in posterior lobe formation causes decreased FGF signaling, which secondarily decreases the growth of Rathke's pouch.

Figure 5.

(A and B) Loss of OTX2 in the ventral diencephalon disrupts FGF signaling. (A) In situ hybridization at e12.5 on sagittal sections shows that Fgf10 is greatly decreased in the ventral diencephalon of Otx2^FX;Nkx2.1-cre mutants (bottom panel) compared with control (top panel). Immunohistochemistry for pERK expression at e11.5 reveals that pERK signaling is reduced in the Otx2^FX;Nkx2.1-cre mutants (bottom panels) compared with control (top panels). Immunohistochemistry for TLE4 at e11.5 shows a slight decrease in expression in mutants (bottom panels) when compared with control (top panels). The photos in first panel were taken at ×100. Scale bar: 100 μm. The photos in the middle and last panels were taken at ×200. Scale bar: 50 μm. (B) In situ hybridization for Six6 at e12.5 and Nkx2.1 at e11.5 revealed no difference between mutant (bottom panels) and control pituitaries (top panels). Immunohistochemistry for SHH at e11.5 suggests there is no difference in expression between mutant (bottom panel) and control (top panel) pituitaries. The photos in the first and last panels were taken at. Scale bar: 50 μm. The photos in the middle panel were taken at. Scale bar: 100 μm. D—dorsal, C—caudal, V—ventral, R—rostral.

Disrupting a single signaling pathway can impact the expression of other signaling pathways and affect patterning of the ventral diencephalon. Loss of expression of genes, such as Noggin, Tcf7l2 and Wnt5a, in the dorsal aspect of the ventral diencephalon where FGF is normally expressed, can permit expansion of the expression domains of genes typically restricted to the rostral aspect (43–45). We examined the expression domains of the transcription factor Six6 (Fig. 5B) and the signaling by sonic hedgehog (SHH, Fig. 5B) by in situ hybridization and immunostaining, respectively. The expression pattern of both genes appeared to obey the molecular boundary between the dorsal-caudal and rostral-ventral areas of the ventral diencephalon. Nkx2.1 is typically expressed throughout the ventral diencephalon at e11.5, and the Otx2^FX;Nkx2.1-cre mutants had no disruption in expression detected by in situ hybridization (Fig. 5B). TLE4 is normally expressed in two domains of the ventral diencephalon: in the dorsal-caudal area ending at the rostral, ventral aspect of the posterior lobe, and in the area rostral and ventral to Rathke's pouch, but only on the inner area of the neural ectoderm in this region. The outer area of the neural ectoderm, at the ventricular zone, is devoid of Tle4 expression. Immunohistochemistry for TLE4 revealed normal boundaries of expression in the Otx2^FX;Nkx2.1-cre mutant, despite the lack of invagination of the neural ectoderm (Fig. 5A) (41,46).

Mutants lack FGF10 signaling and have altered cell proliferation

FGF10 positively regulates anterior pituitary gland growth and cell survival (32,46–49). The reduced amount of pituitary tissue in Otx2^FX;Nkx2.1cre mutants could arise from reduced cell proliferation and/or enhanced cell death. Timed pregnant mice were injected with the nucleoside analog, bromo-deoxy uridine (BrdU), and after 2 h, the embryos were collected and processed. Immunohistochemistry for BrdU incorporation during the S-phase of the cell cycle is a proxy for cell proliferation. Normal pituitaries have BrdU immunopositive, proliferating cells all along the ventral diencephalon at e10.5 and e12.5 (marked with white bars), except for the invaginated region that will become the infundibulum (Fig. 6A) (49). BrdU immuno-positive, proliferating cells are normally present throughout Rathke's pouch, except for the most ventral area where cells are beginning to differentiate at e12.5. At e12.5, the BrdU-labeled cells in Otx2^FX;Nkx2.1-cre mutant pituitaries exhibit an abnormal pattern in the ventral diencephalon, specifically. There are no BrdU-negative cells in Otx2^FX;Nkx2.1-cre mutants at the site where invagination of the ventral diencephalon normally occurs, which is outlined in white (Fig. 6A). At e12.5, the Otx2^FX;Nkx2.1-cre mutant pituitaries have a patch of BrdU-negative cells in the ventral aspect of Rathke's pouch, similar to the pattern in normal mice.

Figure 6.

(A and B) Cell proliferation is affected in the Otx2^FX;Nkx2.1-cre mutant pituitaries. Timed pregnant female mice were injected with BrdU, and embryos were collected after 2 h. (A) Immunohistochemistry was used at e10.5 and e12.5 to detect BrdU-labeled cells. Three controls and three Otx2^FX;Nkx2.1-cre mutants were used for both e10.5 and e12.5. BrdU-labeled cells were counted in Rathke's pouch and in the white marked region of the ventral diencephalon at both e10.5 and e12.5. The region outlined in white in the top right panel is the evaginating ventral diencephalon that will become the posterior lobe in the control pituitary. The photos in the right panel were taken at. Scale bar: 50 μm. The photos in the left panel were taken at. Scale bar: 100 μm. (B) ImageJ IJ 1.46r software with the Cell Counter plug-in from the NIH was used to count the total number of proliferating cells and the Student's T-Test was used to determine significance. There is a significant decrease in proliferating cells in Rathke's pouch at e12.5 in the Otx2^FX;Nkx2.1-cre mutants compared with controls (*P-value < 0.05) and a significant increase in proliferating cells in the ventral diencephalon at e12.5 in the Otx2^FX;Nkx2.1-cre mutant compared with control (**P-value <0.01). There was no significant difference in proliferating cells at e10.5 in either Rathke's pouch or the ventral diencephalon.

In order to quantify cell proliferation, we calculated the percent of DAPI-stained nuclei that were positive for BrdU immunostaining in specific regions of the pituitary at e10.5 and e12.5 in Otx2^FX;Nkx2.1-cre mutants and controls. At e10.5, there is no significant difference in the average percentage of proliferating cells in the ventral diencephalon (Fig. 6A and B). At e12.5, however, the Otx2^FX;Nkx2.1-cre mutant ventral dienephalon displays a significant increase in the number of proliferating cells (Fig. 6B). The percentage of BrdU-positive cells in Rathke's pouch is significantly decreased at e12.5 in the Otx2^FX;Nkx2.1-cre mutants relative to controls (Fig. 6B). The TUNEL assay was performed at e10.5, e12.5 and e14.5, and no differences in cell death were observed in either the ventral diencephalon or Rathke's pouch of Otx2^FX;Nkx2.1-cre mutants (Supplementary Material, Fig. S2).

Loss of OTX2 in the ventral diencephalon does not affect cell specification or hypothalamic–pituitary axis function

To test whether the pituitary cells in Otx2^FX;Nkx2.1-cre mutants were correctly specified, immunohistochemistry was performed at P10 for the hormones TSHβ, ACTH, LHβ, GH, αMSH (Fig. 7A) and FSHβ (data not shown). Immunostaining indicated that all cell types are present, and other than the smaller posterior and anterior lobes in the mutants, there were no obvious differences between the controls and the Otx2^FX;Nkx2.1-cre mutants (n = 3). At e14.5, the anterior lobe of the Otx2^FX;Nkx2.1-cre mutants exhibit normal immunostaining for POU1F1 and αGSU (Supplementary Material, Fig. S3). Immunostaining for arginine vasopressin (AVP), a neurohypophysial hormone, indicates normal projection of AVP neurons from the hypothalamus to the posterior lobe of the pituitary at P2 in the Otx2^FX;Nkx2.1-cre mutants and controls (Fig. 7A).

Figure 7.

(A) Loss of OTX2 in the ventral diencephalon does not affect cell specification in the anterior pituitary lobe. Immunohistochemistry at P10 on coronal sections for the hormones TSHβ, ACTH, LHβ and GH demonstrate no difference between control and the Otx2^FX;Nkx2.1-cre mutants. Immunohistochemistry on coronal sections at P2 for AVP expression in the posterior lobe reveals no difference between control and the Otx2^FX;Nkx2.1-cre mutants. All photos were taken at ×100. Scale bar: 100 μm. (B) The recovering, mutant infundibulum expresses Fgf10. In situ hybridization at e14.5 and e16.5 using coronal sections shows Fg10 expression in the control infundibulum (left panel). Variable expression on Fgf10 is also detected in the small infundibulum tissue of the Otx2^FX;Nkx.1-cre mutants at e14.5 and e16.5 (arrows, middle and right panels). The photos in the top panel were taken at. Scale bar: 50 μm. The photos in the bottom panel were taken at. Scale bar: 100 μm.

We hypothesized that pituitary glands of the Otx2^FX;Nkx2.1-cre mutants were recovering their growth by a resurgence of FGF signaling. Because Fgf10 expression is no longer detectable in the posterior lobe at birth (data not shown), we examined expression between e14.5 and e16.5 by in situ hybridization in coronal pituitary sections of Otx2^FX;Nkx2.1-cre and control mice. We detected robust Fgf10 expression in the posterior lobe of control mice at these times. Variable amounts of Fgf10 expression are detectable in the small, preliminary nubbins of posterior lobe in the Otx2^FX;Nkx2.1-cre mutants (Fig. 7B).

Homozygous mutant mice are viable and present in expected Mendelian ratios (N = 43, P > 0.25). Surprisingly, there is no obvious difference in growth rate or final adult size of mice at 12 weeks of age. We conducted a fertility study by mating Otx2^FX;Nkx2.1-cre mutant females and males to normal Otx2^FX/FX or C57BL/6 mice of the opposite sex. We mated three different Otx2^FX;Nkx2.1-cre mutant and three different Otx2^FX/FX control 6-week-old females to three different C57BL/6 adult males in separate cages. All females gave birth to a litter of pups within 5 weeks. All pups survived to weaning age (21 days). We mated three different Otx2^FX; Nkx2.1-cre mutant and three different Otx2^FX/FX control 12-week-old males to C57BL/6 females. All females of these matings produced a litter of healthy pups within 6 weeks. This indicates that the hypothalamic–pituitary axes recovered sufficiently to support normal growth and fertility despite the loss of OTX2 protein in the ventral diencephalon.

DISCUSSION

Patients heterozygous for mutations in OTX2 can exhibit varying degrees of pituitary hypoplasia, ectopic posterior pituitary gland, invisible pituitary stalk, CPHD and IGHD. Similarly, mice heterozygous for OTX2 loss of function can have pituitary hypoplasia, missing or misplaced pituitary glands, and/or pituitary dysmorphology (12,50). The pattern of OTX2 expression in the developing mouse pituitary points to a primary role in the development of the pituitary stalk and posterior lobe. OTX2 is expressed strongly in the developing posterior pituitary lobe, hypothalamus and other specific regions of the brain, but expression is modest and transient in Rathke's pouch. To identify the mechanism of OTX2 action in pituitary development, we used a cre-loxP strategy to selectively delete Otx2 from either the ventral diencephalon, which forms the posterior pituitary lobe, or early head development and Rathke's pouch. We report that disruption of OTX2 in early head development causes a variable dysmorphic pituitary gland phenotype, whereas loss of OTX2 in Rathke's pouch has no effect on cell specification. OTX2 deficiency in the ventral diencephalon has a profound effect on the initial development of the posterior lobe and pituitary stalk. This is associated with reduced and delayed FGF signaling, which secondarily causes anterior lobe hypoplasia. We discovered that FGF signaling becomes active later in gestation, and this may contribute to the recovery of anterior pituitary gland growth in mice with a neural-ectoderm-specific OTX2 disruption. Thus, loss of OTX2 in the neural ectoderm is sufficient to cause severe, but transient, hypopituitarism.

The hormone deficiency and pituitary hypoplasia characteristic of some human patients has been proposed to arise from the inability of the mutant OTX2 to activate HESX1 and POU1F1 appropriately (3–5). In the developing mouse pituitary, however, OTX2 protein is only transiently expressed in Rathke's pouch at e10.5, and it is essentially absent from the anterior lobe when POU1F1 is first detectable (25). As OTX2 and POU1F1 are not expressed in the same cells at the same time, it is not plausible that the patient's hormone deficiency is due to failed activation of Pou1f1. In addition, the cre drivers that we used to disrupt Otx2 in the oral or visceral ectoderm had no effect on Pou1f1 expression. Otx2 and Hesx1 both are expressed in the AVE and later in the anterior neural ectoderm. In fact, previous experiments with Otx2^−/− chimeras have shown that Otx2 is required for the initiation of Hesx1 transcription during early forebrain development (51). Temporally regulated deletion of Otx2 during development revealed that this transcription factor has an essential role in head development before e10.5, whereas deletion between e12.5 and e14.5 was compatible with normal morphology but impaired growth, and deletion after e14.5 had no effect on growth (38). Therefore, hypopituitarism most likely originates from early patterning defects during gastrulation (11,19,21,22,52).

We employed several cre drivers to assess the significance of the early, transient expression of Otx2 in Rathke's pouch. Removing OTX2 from Rathke's pouch with either Prop1-cre (data not shown) or Pitx2-cre did not result in any abnormalities, suggesting that OTX2 does not play a direct or intrinsic role in anterior pituitary development. Because Pitx2-cre activity is not 100% penetrant in the cells of Rathke's pouch, it is difficult to rule out the possibility that some transient expression of Otx2 occurred. Foxg1-cre activity is reported to be pituitary specific on certain genetic backgrounds and to label essentially every cell in Rathke's pouch (33,34). OTX2 deletion with Foxg1-cre resulted in serious craniofacial defects including anophthalmia, indicating that in this context Foxg1-cre caused Otx2 deficiency in early head development, in addition to the disruption in Rathke's pouch. These fetuses have pituitary dysmorphology but normal cell specification in the anterior lobe. Taken together, the results of these three cre driver experiments support the interpretation that pituitary dysmorphology is secondary to a defect in early head development, rather than a result of OTX2 deficiency in Rathke's pouch.

OTX2 is expressed strongly in the region of the ventral diencephalon that forms the posterior lobe, and this region normally has reduced cell proliferation relative to the areas directly above and below it (25). Disruption of OTX2 in this region profoundly affected posterior lobe development. The cells normally fated to become the posterior lobe failed to leave the cell cycle, evaginate, differentiate or produce FGF10. The anterior lobes of these fetuses exhibited reduced cell proliferation and hypoplasia during early development, despite intact Otx2 expression in Rathke's pouch. This secondary effect is likely due to the lack of FGF signaling from the ventral diencephalon, because disruption of Fgf or Fgfr2 also causes anterior lobe hypoplasia (32,53). We did not observe enhanced cell death in the pituitaries of Otx2^FX;Nkx2.1-cre mutants. This contrasts with the occurrence of increased apoptosis in other neural-derived structures of Otx2 deficient mice, including the retinal pigment epithelium, photoreceptors, GnRH neurons and the choroid plexus (54–56). This also differs with the Fgf10^−/− embryos, which exhibit an agenic pituitary by e15.5 after widespread apoptosis (32). Our mouse model has only a decrease in Fgf10, opposed to a complete loss of Fgf10 throughout the mouse. There are many reports of OTX2 expression affecting FGF signaling, and vice versa. The effects can be positive or negative (57–60). Thus, the relationship between Otx2 and FGF signaling may differ among affected organ systems.

There are many factors besides OTX2 that affect development of the posterior lobe, including LHX2, NKX2.1, RAX, HES1, TCF7L2, SOX3, WNT5A, SHH and TBX3 (61). In humans, TBX3 mutations cause a multiple congenital anomaly known as mammary ulnar syndrome, affecting posterior limbs, mammary and apocrine glands, teeth, puberty and genital development (62–64). In mice, the pattern of Tbx3 expression appears to coincide with Otx2 in the ventral diencephalon, and there is a trace of Tbx3 expression in the ventral aspect of Rathke's pouch. Some effects of Tbx3 deficiency on pituitary development are similar to those we observed in Otx2^FX;Nkx2.1-cre mutants, but the mechanisms appear to be different. Tbx3 knockout mice fail to generate a region of the ventral diencephalon that is negative for SHH, the cells continue to proliferate, fail to invaginate, and there is a modest, transient change in Fgf10 expression. Rathke's pouch is significantly smaller, cell proliferation in the anterior lobe is significantly reduced and expression of the signature transcription factors POU1F1 and TPIT are lost. In contrast, the growth defects of Otx2^FX;Nkx2.1-cre mutant pituitaries are associated with intact SHH patterning and near absence of Fgf10 expression at the normal time of onset. The Tbx3-mutant mice die owing to cardiovascular defects after e14.5, so the recovery of infundibular and anterior pituitary growth cannot be investigated in these mice. Nevertheless, the different mechanisms underlying hypoplasia of the posterior lobe and pituitary stalk support the idea that TBX3 and OTX2 act in separate pathways that contribute to normal development of these structures.

The postnatal recovery of growth and function in both the posterior and anterior lobes of Otx2^FX;Nkx2.1cre mutants is surprising, and it suggests the possibility of compensation for Otx2 deficiency by other genes. FGF signaling may contribute to the recovery because, despite the complete absence of Fgf10 expression and pERK signaling in mutants at e11.5, Fgf10 expression is detectable later in gestation. We also considered the possibility that genes related to Otx2, such as Otx1, are involved in the recovery process. Both the Otx1 and Otx2 genes encode bicoid-like homeodomain proteins that share extensive sequence similarities and have functional overlap in some tissues (21,65).

There are several pieces of evidence that argue against the involvement of Otx1 in the recovery. First, replacement of the Otx2-coding sequences with those of Otx1 does not rescue anterior structures including the eyes, olfactory bulbs, forebrain and pituitary gland (65,66). Second, Otx1 expression in the anterior lobe is initiated postnatally, after recovery has already begun in the Otx2^FX, Nkx2.1-cre mutants (21,65), and we detected no Otx1 expression in the infundibulum or Rathke's pouch during early pituitary development (data not shown). Third, there is no increase in Otx1 transcripts in pituitary glands of Otx2^FX;Nkx2.1-cre mutant mice after birth (data not shown). Thus, OTX1 is less likely to be involved in the recovery than FGF10. The two homeobox genes Emx1 and Emx2 also share regions of Otx2 expression in the forebrain at e10.5 (20). Furthermore, Emx2 and Otx2 interact with each in a dose-dependent manner in diencephalon development (67). It is possible that Emx1 and Emx2 could compensate for the loss of Otx2. There is no increase of Emx1 and Emx2 transcripts in the Otx2^FX;Nkx2.1-cre mutant pituitaries (data not shown), but we cannot rule out an increase in neural tissues.

There are at least two loci in mice that suppress or enhance the effects of Otx2 deficiency (13), although the underlying genes are not yet known. Disruption of Otx2 in the neural ectoderm might result in a more dramatic phenotype on another genetic background. We analyzed the Otx2^FX; Nkx2.1-cre knockout on a mixed C57BL/6, 129/SvJ background, and while the phenotype was consistent among individuals, it may have been more severe on a pure C57BL/6 genetic background. It is possible that the genes that can suppress the effects of Otx2 deficiency are contributing to the recovery of pituitary growth in the Otx2^FX;Nkx2.1-cre mice postnatally. There may be regions of early OTX2 expression in the neuroectoderm that do not share expression with NKX2.1 and remain intact in our mouse model. We cannot rule out a contribution of residual OTX2 to the recovery of pituitary growth.

Our results suggest mechanisms that underlie hypopituitarism in patients with OTX2 mutations. Although OTX2 deficiency in the ventral diencephalon has no effect on Pou1f1 expression and cell specification or hormone expression, reduction of OTX2 in the neural ectoderm could result in a thin or absent stalk, impairing signaling between the hypothalamus and the pituitary gland. This developmental defect could be persistent if the recovery process is less robust in human pituitary development. The variable ocular, craniofacial and pituitary defects that arose in Otx2^FX;Foxg1-cre mutants are consistent with dosage sensitivity for OTX2 in early head development. An element of chance (stochastic effects) and/or environmental influences (epigenetic effects) could contribute to the variable phenotypes and incomplete penetrance in humans and mice. It is clear, however, that variations in interacting genes that confer enhancing and suppressing effects are important factors (5,10,68). Exome sequencing might identify rare, deleterious variants in genes that enhance the effects of OTX2 mutations and give further insight into the origin of hypopituitarism.

METHODS AND MATERIALS

Mice

All mice were housed in a 12-h light, 12-h dark cycle in ventilated cages with unlimited access to tap water and Purina 5020 chow. All procedures were conducted in accordance with the principles and procedures outlined in the National Institutes of Health Guidelines of the Care and Use of Experimental Animals. Otx2^FX/+ mice were provided by Dr Thomas Lamonerie (Institute of Biology Valrose Université Nice, Lyon Gerland, France). The mice were mated to produce homozygotes (Otx2^FX/FX) and maintained with homozygous matings on the 129/Sv background. The C57BL/6J-Tg(Nkx2-1-cre)2Sand/J (Nkx2.1-cre^+/−) mice were purchased from The Jackson Laboratory (stock number 008661, Bar Harbour, ME, USA) and maintained by mating to C57BL/6J mice from The Jackson Laboratory. The B6.129S4-Gt(ROSA)26Sor^tm1Sor/J mice (R26R^FX/FX) were purchased from The Jackson Laboratory (stock number 003474) and maintained as homozygotes. Otx2^FX/FX mice were mated to Nkx2.1-cre^+/− mice to generate Otx2^FX/+;Nkx2.1-cre^+/− mice. Otx2^FX/FX mice were mated to the R26R^FX/FX reporter mouse to generate Otx2^FX/FX;R26R^FX/FXmice. These mice were mated to Otx2^FX/+;Nkx2.1-cre^+/− mice to generate Otx2^FX/FX;Nkx2.1-cre^+/−; R26R^FX/+mice. The Pitx2^tm4(cre)Jfm (Pitx2-cre) mice were obtained from Dr James F. Martin (University of Texas, Southwestern, Dallas, TX, USA) and maintained by mating to C57BL/6J wild-type mice from Jackson Laboratory. Otx2^FX/FX mice were mated to Pitx2-cre^+/− mice to generate Otx2^FX/+;Pitx2-cre^+/− mice. Otx2^FX/FX;R26R^FX/FXmice were mated to Otx2^FX/+;Pitx2-cre^+/− mice to generate Otx2^FX/FX;Pitx2-cre^+/−;R26R^FX/+mice. The Foxg1^tm1(cre)Skm (Foxg1-cre) mice were obtained from Dr Susan K. McConnell (Stanford University, Stanford, CA, USA) and maintained by mating to wild-type Swiss Webster (CFW) mice from Charles River. Otx2^FX/FX mice were mated to Foxg1-cre^+/− mice to generate Otx2^FX/+;Foxg1-cre^+/− mice. Otx2^FX/FX mice were mated to Otx2^FX/+;Foxg1-cre^+/− mice to generate Otx2^FX/FX;Foxg1-cre^+/− mice.

Genotyping was carried out by PCR amplification of genomic DNA as previously described. Cre-specific primers were 5′ GCATAACCAGTGAAACAGCATTGCTG 3′ and 5′ GGACATGTTCAGGGATCGCCAGGCG 3′ (69); R26R^FX primers were 5′ GGCTTAAAGGCTAACCTGATGTG 3′, 5′ -GCGAAGAGTTTGTCCTCAACC 3′ and 5′ GGAGCGGGAGAAATGGATATG 3′ (69); and Otx2^FX were primers were 5′ GAACAAACGTCCCTGTGGTG 3′ and 5′ GAGCTTCCAGAACGTCGAG 3′ (38). PCR products were visualized with ethidium bromide on a 1.5–2% agarose gel.

Timed pregnancies were produced using natural matings of sexually mature females and males. The morning after conception is designated e0.5, and the day of birth is designated as P1. This method was used to collect embryos at the times of e10.5, e11.5, e12.5, e14.5 and e16.5 and postnatal mice at P2 and P10.

Tissue preparation and histology

Embryos were fixed in 4% formaldehyde in PBS at 4°C for 30 min for e10.5 and e11.5 embryos, for 1 h for e12.5 embryos, for 2 h for e14.5 and e16.5 embryos, and overnight for P2 and P10 mice. The tissue was washed with PBS, dehydrated to 70% ethanol and embedded in a Tissue Tek VIP Paraffin tissue processing machine (Miles Scientific). Six-micrometer-thick sagittal sections were prepared from embryos collected at e10.5 through e16.5, and coronal sections were prepared from some e14.5 embryos and postnatal pups. For hematoxylin and eosin (Sigma–Aldrich) staining, the paraffin-embedded tissue sections were soaked in xylene to remove the paraffin and hydrated by soaking slides in 100% ethanol, 95% ethanol and finally distilled water. The slides were next soaked in hematoxylin for 30 s, rinsed in distilled water, soaked in eosin for 20 s and then rinsed in distilled water. The slides were dehydrated back to xylene and mounted with xylene/permount 1 : 2 (Fisher) mounting media. For frozen sections, tissue was fixed in 4% paraformaldehyde in PBS at 4°C for the times previously described, rinsed in PBS and soaked in 30% sucrose at 4° C until the tissue sank to the bottom of the container. This fixed tissue was embedded in OCT (Tissue-Tek), frozen on dry ice and placed at −80°C prior to sectioning at a thickness of 16 μm. X-gal staining and neutral red staining were performed as previously described (69,70).

Immunohistochemistry and in situ hybridization

Immunostaining for pituitary hormone markers was performed using anti-TSHβ, ACTH, LHβ, FSHβ, GH (1 : 1000, National Hormone and Peptide Program, UCLA Medical Center, Torrance, CA, USA), and anti-α-MSH (AB5087, 1 : 100, Chemicon, Temecula, CA, USA) antibodies, on paraffin sections. Sections were hydrated, incubated for 20 min in 3% H₂O₂ : 50% methanol to block endogenous peroxidases. All slides were placed in normal goat serum block (5% goat serum, 3% BSA and 0.5% Tween-20 in PBS) for 10 min at room temperature. All hormone antibodies were diluted in blocking solution and incubated on the slides overnight at 4°C, except for GH, which was incubated for 1 h at room temperature. When staining for anti-OTX2 (ab21990, 1 : 1000, Abcam, Cambridge, MA, USA), anti-TLE4 (1 : 1000, from Dr Stefano Stefani), anti-pERK (1 : 100, 4370, Cell Signaling), anti-AVP (1 : 500, ab39363, Abcam), SHH (AF464, 1 : 200, R&D Systems), anti-PITX1 (1 : 100, from Dr Jacques Drouin) and anti-TPIT (1 : 200, from Dr Jacques Drouin), paraffin sections were first boiled in 0.01 M citrate for 10 min, followed by 20 min of incubation in 3% H₂O₂ : 50% methanol, and were incubated for 10 min in normal goat serum block, except for SHH slides, which required a 5% normal donkey serum block. Antibodies were incubated overnight at 4°C. The following secondary antibodies were used: biotinylated anti-rabbit IgG (BA-1000, 1 : 200, Vector Laboratories, Burlingame, CA, USA) for anti-TSHβ, ACTH, FSHβ OTX2, TLE4, pERK and AVP; anti-human biotin (1 : 200, ab97223, Abcam) for anti-GH, biotinylated anti-guinea pig IgG for anti-LHβ, biotin-conjugated anti-goat IgG (1 : 200, 305-066-047, Jackson Immunoresearch) for anti-SHH and biotin anti-sheep IgG (1 : 200, 313-065-047, Jackson Immunoresearch) for anti-αMSH. Antibodies were detected using either the tyramide signal amplification (TSA) and fluorescein isothiocynate kit (according to protocol, Perkin–Elmer, Boston, MA, USA) or streptavidin-conjugated Alexa-fluor 488 (1 : 200, S11223, Invitrogen). Cell proliferation detection was performed as described by (71) and using 100 mg BrdU per gram of body weight injected into pregnant mice by intraperitoneal injection 2 h prior to collecting embryos. After processing, tissue sections were boiled in 0.01 m citrate for 10 min, followed by 20 min of incubation in 3% H₂O₂ : 50% methanol and incubated for 1 h in a mouse IgG block. The secondary biotin anti-rat IgG (1 : 200, 711-066-152, Jackson Immunoresearch, Westgrove, PA, USA), and the TSA kit were used for detection. Cell nuclei were stained with DAPI (1 : 200) for 5 min. Slides were mounted with permount mounting medium.

Cell death was detected using the In Situ Cell Death Detection Kit, Fluorescein (Roche), which labels DNA strand breaks with TUNEL technology. Paraffin sections were rehydrated as described earlier and then permealibilzed with 20 μg/ml Proteinase K in 10 mm Tris–HCL, pH 8. The TUNEL labeling was carried out according to the published protocol, and the cell nuclei were stained with DAPI (1 : 200) for 5 min. Cell nuclei were stained with DAPI (1 : 200) for 5 min. Slides were mounted with permount mounting medium.

In situ hybridization

Lori Sussel (Columbia University, NY) provided a mouse Nkx2.1 clone in a pSK+ plasmid. The sequence was linearized with SalI and labeled with T3 polymerase. Bridget Hogan (Duke University) provided a mouse Fgf10 clone in a pKSII+ plasmid. The sequence was linearized with BamHI and labeled with T3 polymerase. The mouse Six6 clone in a pFLCI plasmid was linearized with BamHI and labeled with T7 polymerase. Juan Pedro Martinez-Barbera (University College London, UK) provided a mouse Otx1 clone. The sequence was linearized with EcoR1 and labeled with Sp6. Paul Thomas (University of Adelaide, Australia) provided the mouse Hesx1 clone. The sequence was linearized with BamHI and labeled with T3 polymerase. The probes were diluted 1 : 100 and hybridized at 55°C. The mouse Otx2 clone in a pFLCI plasmid was linearized with SmaI and and labeled with T7 polymerase. The Otx2 in situ probe was diluted 1 : 200 and hybridized at 60°C. All riboprobes were generated and labeled with digoxigenin and precipitated with nitro-blue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate (Roche Molecular Biochemicals, Indianapolis, IN, USA) using previously described methods (72,73). All images were taken with a Leica Leitz DMB microscope and Leica DFC310 FX camera. Images were analyzed with Leica Application Suite Software V2.7 and Adobe Photoshop CS6 V13.0.

Isolation of RNA from pituitaries

Postnatal day 1 pituitaries were collected and stored in RNAlater (Ambion, Austin, TX, USA) at (−20°C). RNAlater was removed, and pituitaries were placed in 300 μl lysis buffer from the RNAqueous 4PCR Kit (AMbion). Pituitaries were homogenized using an Ultra-Turrax T8 homogenizer (IKA, Wilmington, NC, USA). Total RNA was isolated with RNAqueous Micro Kit according to manufacturer's instructions. RNA quantity (determined by A260 value) and quality (determined by A260/A280 ratio) were analyzed with a NanoDropTM 1000 Spectrophotometer (Thermo Scientific, Waltham, MA, USA).

RT-qPCR

Synthesis of cDNA was carried out as described (74). The following primers were designed to amplify Otx1 cDNA across exons 2 and 3: 5′ GCAAAGACTCGCTACCCAGA 3′; 5′ GGTTTTCGTTCCATTCCCGC 3′. The following primers were designed to amplify Emx1 cDNA across exons 2 and 3: 5′ GCGAGCCTTTGAGAAGAATCAC 3′; 5′ CCGATTCTGGAACCACACCTT 3′. The following primers were designed to amplify Emx2 cDNA across exons 2 and 3: 5′ AACCATTACGTGGTGGGAGC 3′; 5′ CGCCTGCTTGGTAGCAATTC 3′. PCR products were visualized with ethidium bromide on a 1.5% agarose gel, and bands were confirmed as Otx1, Emx1 and Emx2 PCR products through standard Sanger sequencing at the University of Michigan Sequencing Core. PCR amplification of the housekeeping gene Hprt served as an internal control for each RNA sample.

Pituitary area

Hematoxylin and Eosin Staining (H&E) was performed at P2. Three wild-types and three mutants were analyzed. Four sections were selected at five section intervals from each sample. ImageJ IJ 1.46r software from The National Institute of Health (NIH) was used to measure the posterior and anterior lobes with the trace tool. Images were converted from pixels to centimeters. Average area was calculated for anterior and posterior lobe, as well as standard of deviations and statistical significance using a Student's T-Test, with Excel Software. The average area and standard deviation was divided by 50 because the original images were taken at ×50 magnification. The area was converted from centimeters to micrometers. Graphs were created with Excel software.

Pituitary size

Relative pituitary size was measured at e14.5, P2 and P10. Sagittal sections were used for e14.5 embryos, and coronal sections were used for P2 and P10 mice. Three controls and three Otx2^FX;Nkx2.1-cre mutants were sectioned with two sections per slide. The number of sections with pituitary gland tissue was counted for each age and genotype. The average number of sections, standard deviations, significance using a Student's T-Test, and graphs were calculated using Excel Software. To calculate percentage, the average number of Otx2^FX;Nkx2.1-cre sections was divided by the average number of control sections for each time point.

BrdU cell proliferation counting

Timed pregnant female mice were injected with BrdU, and after 2 h, embryos were collected. BrdU immunostaining was performed at both e10.5 and e12.5. Images were taken at ×20 magnification for DAPI only, BrdU only and the composite for each embryo. Four wild types and four mutants were analyzed. ImageJ IJ 1.46r software from the NIH was used to calculate the number of proliferating cells (BrdU-labeled cells) compared with the total number of DAPI cells. Cells were counted in the region of the ventral diencephalon (as marked on BrdU figure) as well as in Rathke's pouch. The counting was done by using the Cell Counter plug-in available from the Image J software. The average number of cells, standard deviations, significance using a Student's T-Test and graphs were calculated using Excel Software.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work was supported by the National Institute of Heath (grant number R01HD30428 SAC).